Adversarial Attacks on Robot Localization Systems via Deep Feature Perturbation

Zhenyu Li, Tianyi Shang · 2026

Read Adversarial Attacks on Robot Localization Systems via Deep Feature Perturbation by Zhenyu Li, Tianyi Shang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-01892.txt.

Description

Li and Shang attack robot localization through its deep features. Localization guides safe, autonomous navigation for robots. Crafted perturbations can mislead these systems badly. Errors can cause wrong routes or unsafe contact. Mission-critical settings face the highest stakes here.

Their method crafts adversarial queries for visual localization. It targets product quantization inside the retrieval pipeline. A lightweight network perturbs query feature encodings. Fooled retrieval returns irrelevant database entries instead. Queries are built in a forward then backward pass. The backward pass refines the perturbation by optimization. The design stays light in compute cost. Yet the perturbations remain subtle and effective, they report.

Catalog record

CREATORZhenyu Li, Tianyi Shang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.01892

Browse the full library · About the Tower of Babel